How does the RAID Controller battery recalibration process work?

Issue

What is the expected process and approximate time it takes for a Battery Backup Unit re-calibration or initial learn cycle for a replacement Battery Backup Unit on a RAID Controller?

Solution

Detailed below are the specifications and Learn Cycle definitions for the 61xx, 7 and 8 Series Appliance Platforms.

8-Series Platforms:

Models that use a RAID Controller and the RAID Controller Type:

RG280 - DELL PERC H710

M|RG800 - DELL PERC H710

M|RG8000 - DELL PERC H710

S800/8000 - DELL PERC H710

M8800 -  DELL PERC H710

Specifications:

RAID Controller Processor: LSI SAS2208 Dual-core ROC  800MHz

Battery Backup Type: Lithium Ion

Battery Backup Model: LSIiBBU09

Battery Learn Cycle:

• Learn cycle is a battery calibration operation performed by the controller periodically to determine the condition of the battery.
• Controller firmware automatically initiates the learn-cycle every 90 days and cannot be stopped.
• Learn cycle fully discharges and charges the RAID controller battery.
• This operation calculates the charge that is remaining in the battery to ensure there is sufficient capacity.
• This operation causes no impact to system or controller.
• Controller will change to write-through cache policy until the battery learn cycle completes.
• Expected time frame for completion of a learn cycle is about 7 hours and it’ll shorten as the battery capacity deteriorates over time.
• Discharge cycle - 3 hours
• Charge cycle - 4 hours
• Controller will change to write-through cache policy until the battery learn cycle completes.

7-Series Platforms:

Models that use a RAID Controller and the RAID Controller Type:

M|RG700 - DELL PERC H700

M|RG7000 - DELL PERC H700

M7700 - DELL PERC H700

S7000 - DELL PERC H700

Specifications:

RAID Controller Processor: LSI SAS2108 6Gb/s ROC – 800MHz

Battery Backup Type: Lithium Ion

Battery Backup Model: LSIiBBU07

Battery Learn Cycle:

• Learn cycle is a battery calibration operation performed by the controller periodically to determine the condition of the battery.
• Controller firmware automatically initiates the learn cycle every 90 days and cannot be stopped.

The learning mechanism in the battery is required for accounting the changes in battery impedance and no-load chemical full capacity (Qmax) due to battery aging. Depending on the configuration option, the Firmware module periodically performs the re-learn on batteries, for more accurate estimation of the battery capacity.

Qualified learn cycle is the discharge cycle, that might have drained sufficient amount of current/charge that would be sufficient to calibrate the battery capacity with respect to the current, battery Voltage, and the amount of charge drained out etc. Every re-learn updates the full charge capacity (FCC) & max_error registers in the battery gas gauge EEPROM file.

Without a prior qualified learn cycle, for a fresh battery, the FCC register contains, just the value that is being loaded with default gas gauge EEPROM file, which is typically 80% of the Design Capacity. This value just acts as some reference point to the coulomb counter before the calibration cycle.

The learn procedure is as follows:

1. For a freshly inserted battery, or for monthly scheduled re-learns, RAID Controller firmware ensures the battery gets completely charged before initiating a re-learn.
2. Enable charging, and if the battery is charging (positive current), generate an internal log event,†Battery started chargingâ€. The viewable CLI statistic RAID.BATTERYSTATUS will indicate "Charging".
3. Firmware checks if the battery has completed charging, if not, will wait until charging completes, and generate the internal log event,â€Battery relearn pending: Battery is under chargeâ€.
4. If charging is completed (Zero current), generate the internal log event, “Battery charge completeâ€.The viewable CLI statistic RAID.BATTERYSTATUS will indicate "Charged".
5. Then the Firmware turns off Charging, and if the temperature is low enough, enables the discharging Load. And generate an internal log event, Battery relearn startedard.
6. If the battery starts discharging(negative current), generate internal log events, Battery is discharging & the Battery relearn in progress. The viewable CLI statistic RAID.BATTERYSTATUS will indicate "Discharging".
7. When discharging gets completed, generate an internal log event, “Battery relearn completedâ€, and disconnect the discharge load.
8. Turn on charging, and if charging gets kicked-in, generate an internal log event, Battery started charging. The viewable CLI statistic RAID.BATTERYSTATUS will again indicate "Charging".
9. If the max-error does not get reset Back to 2% then the logic automatically generates an internal log event indicating unsuccessful completion of learn by the event, Battery requires reconditioning; please initiate a LEARN cycle. The viewable CLI statistic RAID.BATTERYSTATUS will indicate "Initialize/Recondition Recharging".
10. If EEPROM registers are updated by firmware to improve the battery Life, or for a fresh battery, second re-learn is mostly required, and hence the RAID Controller firmware automatically initiates the second learn cycle. The viewable CLI statistic RAID.BATTERYSTATUS will cycle from "Discharging" to "Charging" to "Charged" as the second relearn completes.

Extra Notes:

Expected time for charging - The battery backup unit charge time should be around 4 hours. During a learn cycle, it takes approximately 7-8 hours to complete as it goes through a series of discharge and charge cycles in order to gauge the capacity of the battery.
Initial charging time is dependent on the condition of the battery. If the initial charge is low (the battery has sat around for a length of time) - the learn cycle is most likely to take much longer.
When a new battery is shipped, its charge level is most likely to drop below 40% after sitting in storage waiting to be purchased. If this is the case, you can expect the learn cycle to be lengthy (upper limit of the 8 hours).
The second re-learn is a requirement actually posed by the gas gauge which is part of battery backup unit hardware. It is the gas gauge that decides if the learn has been a qualified or disqualified one. There are 3 different ways that gas gauge could finish a re-learn. The conditions 2 & 3 require initiation of second re-learn.

1. Qualified : Max Error gets to 2%
2. Disqualified : Max error stays at 100%, Re-learn Requested bit in the gas gauge status register is set
3. Limited : max error gets to 8%, Re-learn Requested bit in the gas gauge status register is set.

The firmware initiates a re-learn. After the discharge cycle, firmware enables the charging cycle. After the battery gets charged to 10% of the full charge capacity, the firmware checks if the re-learn requested bit in the gas gauge status register has been set (which means max  error is not 2%).. if it is set, the firmware initiates the second re-learn cycle.
A Majority of batteries are Typically go for second re-learn, which is normal.
For some batteries, if they have not spent much time on the shelf, they may only need one learn cycle, which is normal.
The Learn cycle usually runs twice for a new battery backup unit, RAID cache is re-enabled during the middle of each learn cycle without obtaining a full charge.
A typical Status result is shown below:

RAID.BATTERYSTATUS Charging
OK Completed

RAID.CACHEON 0
OK Completed

This is as designed. After a re-learn, while charging back to full capacity, Cache Policy turns to Write-Back as soon as the battery capacity goes above the Capacity Threshold for Write-Back.

The LSIiBBU07 battery is designed to have a larger capacity than this threshold.

The battery does not need to be completely charged before supporting the write-back cache policy.
Max error rate can only go as low as 2% for the LSIiBBU07.

61xx Series Platforms:

Models that use a RAID Controller and the RAID Controller Type:

M|RG610 - LSI SAS 9260-8i

M|RG6100 - LSI SAS 9260-8i

Specifications:

RAID Controller Processor: LSI SAS2108 6Gb/s ROC – 800MHz

Battery Backup Type: Lithium Ion

Battery Backup Model: LSIiBBU07

Battery Learn Cycle:

As per the Learn Cycle of the 7-Series outlined above.